The service is still finnicky, but because it still works it's not a top priority.

We believe we've narrowed down the issue to our custom routing software, and we think found a patch for the instability. We're testing the patch before we deploy to production.

We've narrowed down the source of instability to the node, not the routing machine.

Our Eaglercraft server infrastructure is being sensitive right now, our engineers are working on fixing the issue.